This is an author-deposited version published in:
http://oatao.univ-toulouse.fr/
Eprints ID: 9056
To link to this article: DOI:
10.1016/j.apsusc.2012.03.053
URL:
http://dx.doi.org/10.1016/j.apsusc.2012.03.053
To cite this version:
Aloui, Lyacine and Duguet, Thomas and Haidara,
Fanta and Record, Marie-Christine and Samélor, Diane and Senocq,
François and Mangelinck, Dominique and Vahlas, Constantin Al–Cu
intermetallic coatings processed by sequential metalorganic chemical
vapour deposition and post-deposition annealing. (2012) Applied Surface
Science, vol. 258 (n° 17). pp. 6425-6430. ISSN 0169-4332!
O
pen
A
rchive
T
oulouse
A
rchive
O
uverte (
OATAO
)
OATAO is an open access repository that collects the work of Toulouse researchers and
makes it freely available over the web where possible.
Any correspondence concerning this service should be sent to the repository
administrator:
staff-oatao@listes-diff.inp-toulouse.fr
!
Al–Cu
intermetallic
coatings
processed
by
sequential
metalorganic
chemical
vapour
deposition
and
post-deposition
annealing
Lyacine
Aloui
a,
Thomas
Duguet
a,∗,
Fanta
Haidara
b,
Marie-Christine
Record
b,
Diane
Samélor
a,
Franc¸
ois
Senocq
a,
Dominique
Mangelinck
b,
Constantin
Vahlas
aaCIRIMAT,UniversitédeToulouse-CNRS,4alléeEmileMonso,BP-44362,31432ToulouseCedex4,France bIM2NP,UniversitéPaulCézanne-CNRS,FacultédeStJérôme,Case142,13397MarseilleCedex20,France
Keywords: Aluminides Intermetallics
Chemicalvapourdeposition Coatings
Annealing
a
b
s
t
r
a
c
t
Sequentialprocessingofaluminumandcopperfollowedbyreactivediffusionannealingisusedasa paradigmforthemetalorganicchemicalvapourdeposition(MOCVD)ofcoatingscontainingintermetallic alloys.DimethylethylaminealaneandcopperN,N′-di-isopropylacetamidinateareusedasaluminumand
copperprecursors,respectively.Depositionisperformedonsteelandsilicasubstratesat1.33kPaand 493–513K.DifferentoverallcompositionsintheentirerangeoftheAl–Cuphasediagramareobtained byvaryingtherelativethicknessofthetwoelementallayerswhilemaintainingtheoverallthicknessof thecoatingcloseto1mm.As-depositedfilmspresentaroughmorphologyattributedtothedifficulty ofcoppertonucleateonaluminum.Post-depositionannealingismonitoredbyinsituX-raydiffraction, andallowssmootheningthemicrostructureandidentifyingconditionsleadingtoseveralAl–Cuphases. OurresultsestablishaproofofprinciplefollowingwhichMOCVDofmetallicalloysisfeasible,andare expectedtoextendthematerialspoolfornumerousapplications,withinnovativethinfilmprocessing on,andsurfacepropertiesofcomplexinshapeparts.
1. Introduction
Al–Cuintermetalliccompounds present attractiveproperties in applicationssuchas interconnects for integrated circuits [1]
orcorrosionresistantcoatings[2]. Al–Al2Cucompositesfeature
enhancedYoung’smodulus,goodcompressivestrengthand rea-sonablygoodcompressiveductility[3].Inaddition,AlandCuare reasonablyinexpensive,andeasilyavailable.
ProcessingofAl–Cuand,moregenerallyofintermetallicalloy coatingsby metalorganic chemical vapourdeposition (MOCVD) isexpectedtoextendtheirimplementationinsurface engineer-ing. Especially, thanks to the possibility to operate in surface reactioncontrolledregime,MOCVDallows surfacetreatmentof complex-in-shapeitemssuchasglassmoulds,turbinebladesand vanesinaeronauticindustries,orporouspreformswhoseinternal surfacemaybefunctionalizedforthepreparationofsupported cat-alysts.Versatility,costeffectiveness,environmentalcompatibility, andthepossibilitytoprocessfilmscontainingthermodynamically metastablephases,areadditionaladvantagesofMOCVDprocesses. Finally,theuseofmolecularprecursorsallowsoperatingatlowto
∗ Correspondingauthor.Tel.:+33534323439,fax:+33534323498. E-mailaddress:thomas.duguet@ensiacet.fr(T.Duguet).
moderatetemperatures,thusextendingthetargetedapplications spectrumsoastocovertemperature-sensitivesubstrates.
Thepricetopayfor thishighpotentialis theneedtotackle the challenges imposed by thecomplex gas phaseand surface chemistries.Inadditiontomasteringthedepositionreaction,these challenges also concernthe designof the precursors upstream theMOCVDprocess,theengineeringoftheMOCVDapparatusin termsofprecursorvapourgeneration,energydeliverymeans,and dynamicalinsituandonlinediagnosticstomonitorgasand sur-facereactions.Theinherentdifficultyfor theestablishmentofa robustMOCVDprocessisfurtheramplifiedinthecaseofcoatings containingseveralelementsandpotentiallyintermetallicphases, mainlybecauseofthelimitedwidthoftheirstabilitydomains,and thefar-from-equilibriuminitialstatewhichcanleadtounpredicted transitions[4].Moreover,suchaprocessforthepreparationof mul-timetalliccoatingsmustinvolvetheuseofcompatibleprecursors forthedepositedelements.Thegeneralcriteriaqualifyingan inor-ganicormolecularcompoundasprecursorforCVDprocesseswere discussedbyMauryetal.[5,6].InthecaseoftheMOCVDof inter-metalliccompoundsthereareadditionalonessuchas(a)similar transport behaviours,(b)absenceof heteroatomsin theligands whichmayreactwiththeothermetal,(c)compatible decompo-sitionschemes,andifpossible(d)belongingtoacommonfamily ofcompounds.Untilnow,thissituationresultedinlimited investi-gationofMOCVDfortheco-depositionofintermetallicalloyfilms.
ReportsontheMOCVDofAl–Cuwerepublishedin thenineties involvingaluminumalaneandcopperphosphineprecursors[7,8]. However,theyconcernfilmswithlowcopperconcentration,atthe levelof1wt.%,targetingthedopingofAl-basedinterconnectionsin microelectronicsratherthantheprocessingofcoatingscontaining intermetallicphases.
Theabovementionedconstraintscanbepartiallycircumvented ifthepreparationofthecoatingproceedsintwosteps.Namely, sequentialdepositionoftheelementsintheformofbi-or multi-layers,isfollowedbyanappropriatelytunedannealingwhichleads totheformationofthetargetedphases.Thislattersolution,applied intheprocessingofAl/Cubilayers,isadoptedinthepresentstudy. Optimizationofthedepositionreactor,processingoftheAlandCu unaryfilms,investigationoftheirdecompositionmechanisms,and kineticmodellingoftheMOCVDprocessarepresentedelsewhere ([9,10]andreferencestherein).
The article is presented asfollows. Theexperimental proto-colinvolvingMOCVDofAlandCu,andpost-depositionannealing ispresentedin details,first.Then,themicrostructureand com-positionprofilesoftheas-processedAl-richandCu-richbilayers arecompared.Finally,microstructureandphasetransitionsofthe annealedcoatingsarepresentedanddiscussed,priortoproviding concludingremarks.
2. Materialandmethods
Depositionsareperformedintheexperimentalsetupdescribed indetailsandmodeledinRef. [11].Thesetupiscomposedofa stagnantflow,cylindrical,stainless steelreactor.Thedeposition chamberfeaturesadoubleenvelopeallowingthemonitoringof wallstemperaturethroughthecirculationofthermallyregulated silicon oil. A turbomolecular pump ensures a basepressure of 1.3×10−4Pa.Thepumpinggroupisprotectedfromthecorrosive
by-productsbyaliquidnitrogentrap.Gasisdistributedthrough ashowerheadsystem,describedandmodeledinRef.[12].Gases arefedthroughelectropolishedstainlesssteelgaslineswithVCR fittingsandtheirflowrateiscontrolledbycomputerdrivenmass flowcontrollers.
5mm×10mm up to 20mm×20mm 304L stainless steel couponsareusedasrepresentativeoftechnologicallyinteresting substrates.Thermallyoxidizedsilicon(140nmSiO2)couponsare
usedfortheeaseofcrosssectionspreparationforobservationby scanningelectronmicroscopy(SEM),andforX-raydiffraction anal-yses(XRD).Substratesareplacedhorizontallyona58mmdiameter susceptorstandingbelowtheshowerhead.Theyareheatedbya resistancecoilgyredjustbelowthesurfaceofthesusceptor. Stain-lesssteelsubstratesarepolisheddownto4000SiCpapergrade andaresonicatedinacetoneandanhydrousethanol.Siliconwafers aredegreasedina70%H2SO4–30%H2O2solution,rinsedwith
de-ionizedwateranddriedunderargonstreambeforeuse.Asetof fivestainlesssteelandfiveSiO2substratesisusedineachrun.
Sub-stratesareexposedtoatmosphereforalimitedtimeduringtheir transferbetweenthepreparationlabandtheirloadinginto vac-uum.Priortodeposition,insituradiofrequency(RF)Ar–10%H2
plasmaetchingisappliedwithinputpower40Wat120kHz,in conditions160Paand493K,for30minwiththeaimtorecoveran organic-pollution-freesteelorsilicasurface.Inallexperimentsthe operatingpressureandthetemperatureofthereactorwallsare fixedat1.33kPaand368K,respectively.
Adduct gradeDMEAA(SAFC Hitech)is usedas-receivedina stainlesssteel bubbler. It is maintained at 281K by immersion in a thermoregulated water bath. The corresponding saturated vapourpressure is 99Pa. TheDMEAA bubbler is maintainedat thistemperatureduringtheentireperiodofitsserviceinorder to avoid degradation of the precursor [13]. 25 standard cubic
centimetres(sccm)of99.9992%purenitrogen(AirProducts) bub-blesthroughtheAlprecursor.Assumingsaturationofthegasphase, theseconditionsleadtoanupperlimitoftheDMEAAflow rate equalto2sccm[14].Thetotalflowrateiscompletedto327sccm byadding300sccmofN2asadilutiongas.Deposition
tempera-tureisfixedat493K.Ithasbeenpreviouslyshownthatinthese conditionsa meannucleationdelayof7minprecedes initiation ofthegrowthofAl.Nucleationdelayisdeterminedby observa-tionof changeofthecolour of thesurfaceand is thesamefor SiO2 and stainlesssteelsamples.Algrowthrateismeasuredat
certainpositionsoverthesusceptor,throughtheweightgainof eachsample[10].Weightgainispreferredtothickness measure-mentsincefilmporosityandroughnessinduceanoverestimation ofthegrowthrate.Forinstance,SEManalysisofthecross-section ofanAl(resp.Cu)sampleshoweda1300nm(resp.110nm)thick film,whereasthethicknessdeterminedbyweightmeasurement was 400nm (resp. 45nm), only. Al growth rate is mapped as beingconstantat12.6mmolcm−2h−1 onthecentralpartof the
susceptorandgraduallyincreasingbeyondaradiusof15mmto reach15.6mmolcm−2h−1attheedgeofthesusceptor,averaging
13.3mmolcm−2h−1ontheentireheatedsurface.
[Cu(i-Pr-Me-AMD)]2 (NanoMePS, www.nanomeps.fr, last
accessedOctober 7,2011)isusedas-receivedforCudeposition. ThisprecursorisappropriateforuseinaprocessinvolvingCVDof AlfromDMEAAbecause(a)thetwoprocessingconditions win-dowspartiallyoverlap,(b)itcontainsneitheroxygennorhalogens intheligands,noritrequiresoxygencontainingco-reactantsfor thedepositionofcopper[9].[Cu(i-Pr-Me-AMD)]2ismanipulated
in glove box and is conditioned in a packed bed loaded in a homemadesublimatorcomposedofafullstainlesssteelbody,a fritandVCRfittings.Aloadof500mgoffreshcompoundisused in each run. During deposition the precursor is maintained at 368K withthermally regulatedheatingtapes,this temperature correspondingtoasaturatedvapourpressureof36Pa[9].50sccm of N2 are fed through the copper precursor corresponding to
anupperlimit oftheflow rateof[Cu(i-Pr-Me-AMD)]2 equal to
1.2sccm.50sccmof95%purehydrogen(AirProducts)isusedas reducinggas.TherelativelylowpurityofH2doesnotimpactthe
purity ofthe deposited Cu.Similar tothedepositionof Al, the totalflowrateiscompletedto326sccmbyadding225sccmof N2 dilution gas.Cudepositionisperformedat 513K.Themean
growthrateofCuintheseconditionswaspreviouslydetermined byweightgaintobeequalto0.3mmolcm−2h−1[15].
Thedepositionprotocolconsistsin(a)establishingflowratesin allthegaslines,bypassingtheprecursorvessels,(b)establishingthe targetedtemperatureateachpartofthesetupexceptforthe cop-persublimator,(c)performingthedepositionofAl,(d)bypassing theDMEAAbubblerfor30min,(e)heatingthecoppersublimator to368K(10min)andincreasingthetemperatureofthe suscep-torto513K,(f)performingdepositionofCu,and(g)bypassingthe precursorvesselswhilecoolingdownthesusceptor.This proto-colpresentsthedrawbackofmaintainingthefreesurfaceofthe depositedAlduring40min(steps(d)and(e))priorthedeposition ofCu,runningtheriskofcontaminationoftheAl/Cuinterfaceby residualoxygen.However,contaminationlevelislowerthanthe oneobtainedifusingO-containingCuprecursors.
Severaldepositionrunsareperformedinthesameconditions, thedifferencebeingthedurationsofthedepositionofAl(between 20minand57min)andCu(between170minand960min).
Post-depositionannealingisappliedtotheas-processedAl/Cu bilayersinordertoinvestigatereactivediffusion,andobtain coat-ingscontainingdifferentintermetallicAl–Cuphases.InsituXRD measurements in Bragg–Brentano configuration are performed duringheattreatmentsintwoinstruments,operatingwithCuKa, Nifilteredradiation: aBrukerD8Advance,fittedwitha Vantec SuperSpeeddetectorandaPhilipsX’pert.Theyareequippedwith
Fig.1.SurfaceSEMmicrographofanAlfilmdepositedonthermallygrownsilicain theadoptedconditionswithinsituplasmapre-treatment.
aMRI(undervacuum)andanAntonPaarHTTK(undercontrolled atmosphere)hightemperaturechambers,respectively.TheX-ray diffractogramsarerecordedfromroomtemperatureupto928K bystepsof30◦.Crystallographiccharacteristicsoftheas-deposited
andannealedcoatingsaredeterminedbyXRDingrazingincidence usingaSeifertXRD3000TTinstrument(CuKa,graphitediffracted beammonochromator).
Thearithmeticaverageroughness(Ra)ofas-depositedbilayers
andpostannealedcoatingsisdeterminedwithaZygoMetroPro, NewView100opticalprofilometer.Theirmorphologyisevaluated witha LEO 435-VPscanning electron microscope.The elemen-talcompositionofthecoatingsisdeterminedbyElectronProbe Micro-Analysis(EPMA)withaCAMECASX-50apparatus,equipped withthreewavelengthdispersivespectrometers.Depthprofilesare determinedbyradiofrequencyglow dischargeopticalemission spectrometry(RFGD-OES)withaHoribaScientificGD-Profiler2. RFGD-OESuses alow pressure plasmafor fastsputtering(few mm/min)ofthesurfaceofthesampleandexcitationofthe sput-tered atoms. Light emitted by sputtered atoms during species de-excitationiscollectedbyapolychromatorthatisabletodetect allelementsfarfromUVtolowinfrared(includingH,O,CorN species).Theuseofradiofrequencysuppressestheconstraintof usingconductingsamples,henceoxidizedSisubstratescanbe char-acterizedreadily.
3. Resultsanddiscussion 3.1. As-depositedfilms
As-depositedfilmsaresystematicallycomposedofelementalAl andCu.Accordingtodepthprofilingresults,reactivediffusionatthe Al–CuinterfaceisinitiatedduringthedepositionofCu(deposition at513K),butthequantityofintermetallicphasesisnotenoughto revealcorrespondingpeaksintheX-raydiffractograms.The con-tentofheteroatomsduetotheprecursorligandsortotheresidual atmosphereofthereactorvessel(oxygen,nitrogen,carbon)is sys-tematicallybelowthedetectionlimitofEPMA(<1at.%),withour apparatus.
Fig.1presentsaSEMmicrographofAldepositedonthermally grownsilicaafterinsitu plasmapre-treatmentofthesubstrate (Fig.1a)plasmapre-treatmentyieldingasmoothersurface.Ravalue
forthisfilmis0.024mm,tobecomparedwith0.105mmfor unpro-cessedsubstrates.
CudepositiononAlfilmsprovidesbilayerswhosesurface char-acteristicsdependonthequantityperunitsurfaceofthedeposited Cu.SmallquantitiesofdepositedCu,correspondingtoahighAl:Cu
Fig.2.Al-richAl/Cubilayerwithoverallcomposition11at.%Cu.(a)Surface(bottom) andcrosssection(top)SEMmicrographsand(b)RFGD-OESdepthprofileofabilayer formedonsilica.
ratio,do notmarkedlymodifythemorphologyofthefilmwith regardtotheoneshowninFig.1.ASEMcrosssectionanda sur-faceviewofanas-depositedbilayerwithanoverallcomposition of11at.%CuisshowninFig.2a.Thecrosssectionofthefilmis typicalofthatofCVDAl[16].Raofthefilmequals0.087mmand
thesurfacemicrostructureisorganizedintwolevels:anAllayerof approx.1mm,andathinnerCulayerof0.1mm.
Abetterinsightintheorganizationofthebilayerisobtainedby performingtheRFGD-OESdepthprofileanalysis.Fig.2bshowsthe sampleelementalcompositionvs.depth.Fourlayerscanbe iden-tified.Startingfromthesurface,wedeterminethattheCulayer is0.08mm-thick,andthatitcontainsCandO.Cusitsontopofa 0.75mm-thickAllayerthatalsocontainsCandOcontaminants.Itis interestingtonotethatCudiffusesdeepbelowthesurface(approx. downto0.8mm).AsmallCupeakbetween0.65and0.80mmdepth maycorrespondto fastdiffusionof Cuin preferential diffusion pathsand Cusilicidesformation,ortotheabruptchangeofthe sputteringrateattheAl/silicainterface.
CandOpeakspresentattheAl/Cuinterfacemaycorrespond toaslightoxidationoftheAlsurfacebeforeCudepositionstarts. Thethirdlayer(between0.8and1.0mmdepth)correspondstothe thermallyoxidizedSiOxlayer,followedbythepureSiwafer(depth
>1.4mm).
TopographyofoursampleshastobeconsideredforRF GD-OES analysesbecausesputtering occurson alargesurface area (2mmdiameteranode)[17].First,samplespresentarough mor-phologywithmoreorlessdenseregions,porosities,andsuch.This
Fig.3. Cu-richAl/Cubilayerwithoverallcomposition90at.%Cu.(a)Surface (bot-tom)andcrosssection(top)SEMmicrographsofanas-processedsamplegrownon silica.(b)GD-OESdepthprofileofabilayerformedonstainlesssteel.
canresultinthesimultaneoussputteringofthesurfaceandofa deeperfeature,suchasanopenporosity.Therefore,itis impor-tanttocarefullyevaluatecontaminantslevelandsharpinterfaces composition.Withflat,dense,andsmoothsamples,aflat-bottom crateriscreatedwithRFGD-OESthatcorrespondstoa pseudo-layer-by-layeretching.Withimperfectsampleswesuspectthata deeper(next)layermaybesputteredbeforetheendoftheprevious layersputtering.Intheprofileplot,thiswouldcorrespondto visu-alizingtwo(ormore)elementscomingfromdifferentlayersatthe sametimeevenwhereasthechemicalinterfaceisactuallysharp, normaltothesurface.Anotherpointistheabsenceofultra-high vacuumthatmightimplythatO,H,andCcompositionsarelargely overestimatedcomparedtorealvaluesinthefilms.EPMA mea-surementsafterAlandCudepositionsshowcontaminantslevels undertheapparatusdetectionlimit(<1at.%).Additionally, hydro-genisknowntohaveasubstantialeffectontheintensitiesofthe opticalemissionspectra[18].Weignoredthiseffect,anddonot presenthydrogenin thequantitativeprofiles,althoughHsignal wasacquired.Sincethesignalsarenormalizedto100%,neglecting hydrogeninducesanotheruncertaintyontheoverallcomposition. Large quantities of deposited Cu, corresponding to Al–Cu coatingswithhighCu content yieldas-deposited bilayerswith roughsurface.Elementalcompositionofthefollowingsamplesis 90–92at.%Cu,dependingontheirrelativepositionintheMOCVD reactor.Fig.3apresentsacrosssectionandsurfaceSEM micro-graphsofanas-processedfilmonthermallygrownsilica.Thecross sectionrevealsaporousmicrostructurewhichwasdevelopedon athinsublayerofAl.Thisisalsoillustratedonthesurfaceview
ofthesamplewhichpresentsamicrostructurewithRaequal to
0.161mm,typicaltothatofCVDCu[15].Densityfunctional the-ory(DFT)calculationsrevealedthatthefirststageofnucleationof CuonatomicallysmoothsurfaceofAlisnotenergeticallyfavoured. AdsorbedCuadatomsarethusexpectedeithertosegregatebeneath thesurfaceofAlortoformnucleusonsitesofhighenergy,such asothersurfaceadatomsordefects[19].WhennucleationofCu isachieved,additionalCuadatomsareexpectedtodiffuseonthe surfacesoastocontributetothegrowthoftheexistingCugrains. Althoughnottakingintoaccounttheproperreactionschemeofthe MOCVDprocessandtheinitialsurfacestate,thisgrowthmodeis atleastpartiallyresponsibleforthefinalroughmicrostructureof thefilm;i.e.throughtheformationofadiscontinuousfilmwith importantopenporosity.
TheGD-OESelemental profileof anAl/Cubilayerformedon stainlesssteelisshowninFig.3b.Weidentifydifferentlayersfrom thisplot.Theextremesurfacelayer(approx.20nm)correspondsto adsorbedCandOspecies.Then,aCulayerextendsover0.25mm.It seemstocontain10at.%O.Consideringthecommentsmadeabove andbecauseEPMAdoesnotdetectO,weassumethatthe effec-tiveOcontentislower.Thislayerisfollowedbyaslightlyoxidized Al/Cuinterface,asshownbytheOpeakat0.25mm.Thefollowing 0.1mm-thickAllayershowsaveryhighcontentofCuthatconfirms thefastdiffusionofCuatdepositiontemperature(513K).Wedo notknowwhere/howdiffusionoccurs,though.ThefactthatX-ray diffractionresultsshowfcc-Alatlowtemperaturesimpliesthatat mostan(Al)solidsolutionisformed,butobviouslynotwith59at.% Cuasitseemstobeat0.3mmdepth.Again,excitedCu*andAl* atomsco-existintheGDplasmabecausetheyaresputteredatthe sametime,butnotnecessarilybecausetheyco-existintheAllayer. Finally,theinterfacewithstainlesssteeliscomplex,withpossible interdiffusionofAlandCuintothesubstrateandofFeandalloying elementsintothecoating.
3.2. Post-depositionannealing
Fig. 4 presents two high temperature X-ray diffractograms obtainedforcoatingscontaining19at.%Cu(a)and35at.%Cu(b). Thethicknessofbothsamplesisapprox.1mmafterannealing,as determinedbySEMcrosssectionanalyses.Theirmicrostructureis compactandhomogeneous(nolayersaredistinguishable)showing thatthewholecoatingthicknessisalloyed.XRDintensitylevelsare representedbythegrayscale.Inbothsamples,reactionbetween AlandCuduringheattreatmentresultsintheevolutionoftheAl andCupeaks,andtheappearanceofpeakscorrespondingto inter-metalliccompounds.PeaksofAlandCueitherdisappearorare shifted(Fig.4a).Afterheattreatment,sample(a)iscomposedofa mixtureofsolidsolutiona-(Al),u-Al2Cu,andasmallamountofCu
forwhichapeakisstillvisible.Sample(b)issingle-phased,with onlyh-AlCupeaks.Heattreatmentofsample(a)wasstoppedat 853Kbecauseofthevanishingofthediffractionpeaksatthis tem-perature.Thisisattributedtotheappearanceofaliquidphaseas canbeconcludedbytheeutecticreactionoftheAl–Cuphase dia-gramoccurringat821Kforalloyswithcompositionsbetween(Al) andAl2Cu[20].Thesameconclusionisdrawnforsample(b)where
annealingshouldhaveresultedintheformationofatwo-phase samplecomposedofu-Al2Cuandof12%weightfractionofh-AlCu.
ThevanishingoftheAl2Cupeaksabove823K(550◦CinFig.4b)
maycorrespondtothemeltingofAl2Cu,thatshouldoccurabove
863K.
AccordingtotheinsituX-rayanalysis,AlandCufirstreactto formAl2Cu.IfthecompositionisrichenoughinCu(sampleb),then
Al2CureactswiththeremainingCutoformAlCu,above823K.For
sample(b),wesuspectthatg-Al4Cu9isalsoformedalongwith
u-Al2Cu.Itsmaindiffractionpeak(330)appearsataboutthesame
Fig.4.X-raydiffractogramsrecordedasafunctionoftemperatureduring post-depositionannealingforsampleswithcomposition19at.%Cu(a)and35at.%Cu (b).
HoweverifAl4Cu9isformed,itischaracteristicofatransientphase
becauseit wouldappear anddisappearduringtheformationof Al2CuwhilebothAlandCuarestillpresent[21].
Fig.5 presentsasurface SEMviewofthesamplecontaining 19at.%Cuafterannealingat853K.Thesurfaceappearsasbeing
Fig.5.LowmagnificationSEMsurfaceviewofafilmwithoverallcomposition 19at.%Cu.NotationscorrespondtoatomicpercentofCu.
non-homogeneouswithatleasttwozones,illustratedbycurved brightstripesandagreybackground.ThevalueofRainthiscase
is0.212mm.EPMAanalysisrevealedthepresenceofthreezones withvariableelementalcomposition,namely5–8,17and30at.% Cu.Althoughtheprobesizeoftheinstrumentis5mm;i.e.larger thanthetypicallengthscaleoftheobservedfeatures,theseresults confirmtheheterogeneityofthesample.Thismicrostructureisdue tothethermalhistoryofthefilmwhichwascooleddownfroma partiallyliquidstatetoroomtemperature.Thecompositionofthe threezonesofthefilmfitstheeutecticequilibriumattheAl-rich sideoftheAl–Cuphasediagram,involving3at.%Cusolidsolution fcca-(Al),u-Al2Cuandaliquidphaseat17at.%Cu.Thepresence
inthefilmofaphasewhosecompositioncorrespondstotheliquid phaseisattributedtotherelativelyhighcoolingrateof30◦/min.
Basedontheseremarks,elementalcompositionswerenotedinthe threephasesofthemicrographinFig.5.Theultimate microstruc-tureofsuchcoatingsiscompact.Itconfirmsthedeterminedoverall compositionofthecoating,andillustratesthecoherencebetween theoverallcompositionofthecoatingandthecorrespondingphase equilibrium,atleastinthisregionofthephasediagram.
4. Conclusions
Bilayers ofCuand Alweredeposited byMOCVDat 1.33kPa and temperatures 493K and 513K on stainless steel and oxidized Si substrates with the aim to process Al–Cu inter-metallic alloy films. Dimethylethylamine alane and copper N,N′-di-isopropylacetamidinate provided pure Al and Cu films,
respectively.Alfilmsaresmooth,whereassubsequentlydeposited Cu yields surfaces whose roughness increases with increasing thickness.Reactivediffusionisobservedthroughpost-deposition annealing,inducingtheformationofAl–Cuintermetallicphases. InsituX-raydiffractionmeasurementsduringheattreatmentallow investigationofthereactionpath.Depositionofdifferentrelative thicknessesofCuandAlfilmsleadtodifferentsample composi-tions,andallowstheexplorationofdifferentphasespacesofthe Al–Cuphasediagram.Filmscontainingu-Al2Cu,h-AlCu,andlikely
g-Al4Cu9areobtained.
Forthcoming publications will focus on the investigation of thesampleswithanintermediateCucontent,andofappropriate annealingconditions,withtheaimofcreatingsingle-phase inter-metalliccoatingsoftechnologicalinterest(g-Al4Cu9 orz-Al3Cu4,
forinstance).
TheobtainedresultsshowthatMOCVDassociatedwith post-deposition heat treatment is a valid way to obtain films of intermetallicalloys,pavingthewaytoconformaldepositionofthis typeofmaterialsfornumerousapplicationfields.
Acknowledgements
We are indebted to Sophie Gouy and Philippe de Parseval, ObservatoireMidi-Pyrénées,Toulouse,andtoCéliaOlivero,Horiba Jobin-YvonSAS,forEPMAandRF-GD-OESanalyses,respectively. Thisworkwassupportedbythe6thFrameworkEUNetworkof Excellence‘ComplexMetallicAlloys’(contractno. NMP3-CT-2005-500140),andbytheFrenchAgenceNationaledelaRecherche(ANR) undercontractno.NT05-341834.
References
[1]P.-P.Choi,T.Al-Kassab,R.Kirchheim,ScriptaMater.53(2005)323–327. [2]W.R.Osorio,J.E.Spinelli,C.M.A.Freire,M.B.Cardona,A.Garcia,J.AlloysCompd.
443(2007)87–93.
[3]C.J.Hsu,P.W.Kao,N.J.Ho,ScriptaMater.53(2005)341–345.
[4] C.Vahlas,Chemicalvapordepositionofmetals:fromunarysystemstocomplex metallicalloys,in:E.Belin-Ferré(Ed.),ComplexMetallicAlloys:Surfacesand Coatings,WorldScientific,Singapore,2010,pp.49–81.
[5]F.Maury,J.dePhysiqueIVC5(1995)449–463.
[6] F.Maury,L.Gueroudji,C.Vahlas,Surf.Coat.Technol.86–87(1996)316–324. [7]T.Katagiri,E.Kondoh,N.Takeyasu,T.Nakano,H.Yamamoto,T.Otha,Jpn.J.
Appl.Phys.,Part232(1993)L1078–L1080.
[8]E.Kondoh,Y.Kawano,N.Takeyasu,T.Ohta,J.Electrochem.Soc.141(1994) 3494–3499.
[9]V.V.Krisyuk,L.Aloui,N.Prud’homme,S.Sysoev,F.Senocq,D.Samélor,C.Vahlas, Electrochem.Solid-StateLett.14(2011)D26–D29.
[10]T.C.Xenidou,N.Prud’homme,C.Vahlas,N.C.Markatos,A.G.Boudouvis,J. Elec-trochem.Soc.157(2010)D633–D641.
[11]T.C.Xenidou,A.G.Boudouvis,N.C.Markatos,D.Samélor,F.Senocq,A.N.Gleizes, N.Prud’homme,C.Vahlas,Surf.Coat.Technol.201(2007)8868–8872. [12]T.C.Xenidou,A.G.Boudouvis,N.C.Markatos,N.Prud’homme,L.Aloui,C.Vahlas,
ECSTrans.25(2009)1053–1060.
[13]H.Matsuhashi,C.-H.Lee,T.Nishimura,K.Masu,K.Tsubouchi,Mater.Sci. Semi-cond.Process.2(1999)303–308.
[14]C.Vahlas,B.Caussat,F.Senocq,W.L.Gladfelter,L.Aloui,T.Moersch,Chem.Vap. Deposition13(2007)123–129.
[15]V.V.Krisuyk,L.Aloui,N.Prud’homme,B.Sarapata,F.Senocq,D.Samélor,C. Vahlas,ECSTrans.25(2009)581–586.
[16]C.Vahlas,P.Ortiz,D.Oquab,I.W.Hall,J.Electrochem.Soc.148(2001)583–589. [17] N.Trigoulet,T.Hashimoto,S.Molchan,P.Skeldon,G.E.Thompson,A.Tempez,
B.Chapon,Surf.InterfaceAnal.42(2010)328–333.
[18] R.Payling,M.Aeberhard,D.Delfosse,J.Anal.At.Spectrom.16(2001)50–55. [19]A.Benali,C.Lacaze-Dufaure,J.Morillo,Surf.Sci.605(2011)341–350. [20] J.L.Murray,Int.MetalsRev.30(1985)211.